Abstract
It is known that human observers have a preferred initial fixation location on the face which enables efficient integration of features and maximizes performance across a variety of face recognition tasks (Peterson & Eckstein, 2012). Less is known about the influence of fixating the preferred point of fixation on face part judgments. We hypothesize that fixating the preferred point is detrimental relative to other points of fixation in a task that requires making judgments on parts of faces. Fourteen observers completed a face identification task with 4 faces (faces A and B, and top-bottom composite faces AB and BA) embedded in luminance noise. We first used a free eye movements task to measure the preferred initial point of fixation on the face for each individual. Subsequently, in a parts task, observers judged the top or bottom half (based on the block) as matched or unmatched across two sequentially presented faces (Composite face effect, CFE, 50% matched, 50% unmatched, 200 msec. presentations). Within each block, observers were shown either aligned or misaligned faces, and they maintained their gaze either at their preferred point of fixation or 5.2° below it. When judging the top halves of faces while maintaining gaze at their preferred fixation point, observers performed worse in the aligned vs. the misaligned condition (PCaligned = 63.3%, PCmisaligned = 73%, pt-test = 0.005). This is the classic CFE (Young, 1987). This effect disappeared when observers were forced to fixate at a non-preferred location (PCaligned = 62.1%, PCmisaligned = 64.2%, pt-test = 0.34). A control condition judging bottom halves discounted an explanation in terms of spatial proximity of fixation to the judged part (bottom non-preferred PCaligned = 90.2%, PCmisaligned = 89.5%, pt-test = 0.39). The results suggest that face processing at the preferred point of fixation might be mediated by distinct computations that optimize integration of facial features but impair parts judgments.